1. Field of the Invention
The present invention relates to a system and a method for allocating a bitstream of scalable video coding (SVC).
2. Description of Related Art
In the development stream of the video coding, currently the SVC is considered as the video compression standard for the next age. Meanwhile, the SVC is also considered as the first video compression standard today to make the SVC framework normative. It is curtained that the video compression technology proceeds towards a hierarchical scalability and bitstream scalability in future. Once various future video compression standards with the scalability feature were launched, it would be seen that a reasonable mechanism capable of assigning bitstream between each layer within each video compression standard is the vital key for the standard to be survived.
In terms of the current SVC standards, after an extractor extracts a compressed bitstream, three types of bitstream amounts can be adjusted and accommodated: frame rate (temporal scalability), frame resolution (spatial scalability) and frame quality (SNR scalability, i.e., signal-to-noise rate scalability). In addition, the above-mentioned three types of bitstream amounts are allocated in uniform mode, which means the bitstream with the three properties is extracted in bracketed cutting-off manner to achieve the goal of lowering the required bitstream amount according to the conventional SVC standard. However, the above-mentioned approach is disadvantageous in causing an unstable image quality and reducing the bandwidth usage efficiency, since a bracketed cutting-off manner applied to each frame with different complexities would degrade the quality of a frame with high complexity, while enhance the quality of a frame with low complexity not significantly which is equivalent to wasting bitstream amount, nevertheless.
Based on the above-described situation, I. Amonou et al. disclosed a new method in the paper “Optimized rate-distortion extraction with quality layers,” (I. Amonou, N. Cammas, S. Kervadec, S. Pateux, The Image Processing of IEEE International Conference, October 2006). With the method proposed by I. Amonou et al, the scalability of a quality layer is planned by using the R-D optimization (rate-distortion optimization). In other words, the optimum solution is obtained by adjusting the SNR of a bitstream and using Lagranian algorithm, wherein different bandwidths are chosen to extract the SNR characteristics in the bitstream according to the total bandwidth and the complexities of different frames therein, so as to achieve better frame quality.
However, the paper only focuses on the quality layer of a video bitstream for adjusting without considering the adjustments and allocations regarding temporal performance and spatial performance; plus, the paper does not involve the adjustment and allocation in the case where the three parameters (temporal parameter, spatial parameter and SNR parameter) are dynamically and mutually used.